Statistical Technique Cons

Question 1

Chi-squared test

Answer 1

Data must be in frequencies
Must have a precise numerical value that can be organised into categories or groups.
More than 20 observations are required
Time consuming and complicated so high chance of error.
Sometimes groups have to be combined to get specific frequencies so results may not be as detailed.

Question 2

Nearest neighbour

Answer 2

Subject had to be carefully defined, Eg what is a Town, Village,Hamlet.
Need to limit area carefully, as to not just miss a location.
distances are straight lines when road distance will be longer.
Need a minimum of 30 points.
Large number of points makes it time consuming and tedious to calculate.
Doesnt always take into account other controlling factors, Eg mountains. random distribution, doesnt mean it is actually random.

Question 3

Spearmans rank

Answer 3

A sample size of at least 7 pairs is required to make the results worthwhile.
inherent inaccuracy due to ranking of data which ignores specific magnitude of differences.
Only test linear relationships and two sets of variable data.
pearsons tends to be more accurate as it uses actual, rather than ranked values.
Time consuming if data set is large. it is also harder to rank values if they are close together or multiple of the same.

Question 4

Linear regression

Answer 4

Linear regression only looks at linear relationships between dependent and independent variables.
Cant be used for non linear relationships.
does not explain correlation.
sensitive to extreme values.
Uses a more complicated maths formula.
Time consuming and complicated.

Question 5

Pearsons product

Answer 5

Longer and more complex calculation than spearmans.
Significance correlation does not always prove casual relationship.
Assumes that data has a normal distribution.
Only tests linear data and two sets of variable data
Can be affected by extreme values